Cranor, Touretzky Named 2020 AAAS Fellows

Lorrie Cranor and David S. Touretzky, both faculty members in the School of Computer Science, are among almost 500 members of the American Association for the Advancement of Science (AAAS) to be named 2020 AAAS fellows. The lifetime distinction recognizes important contributions to STEM disciplines, including pioneering research, leadership within a given field, teaching and mentoring, fostering collaborations, and advancing public understanding of science. Previous notable recipients include Thomas Edison, Margaret Mead and Grace Hopper. Cranor is the director and Bosch Distinguished Professor in Security and Privacy Technologies of CyLab, and is the FORE Systems Professor of Computer Science and of Engineering and Public Policy in the Institute for Software Research and the Department of Engineering and Public Policy. The AAAS cited her "for contributions to usable privacy and security research, policy and education." She founded the Symposium On Usable Privacy and Security (SOUPS) conference and co-edited the seminal book "Security and Usability." She regularly presents privacy research in Washington, D.C., policy forums and served as chief technologist of the Federal Trade Commission in 2016.

New Technique Isolates Brain Cells Associated With Parkinson's Disease

Analysis Shows Changes in How Diseased Cells Sense and Use Oxygen

Carnegie Mellon University researchers have developed a new technique for isolating a type of brain cell associated with Parkinson's disease symptoms, enabling them to study that cell type in detail. The technique, which works only in specially bred mice, costs less than previous methods for isolating these brain cells, said Alyssa Lawler, a Ph.D. student in biological sciences. By using it, she and her colleagues already have detected previously undiscovered changes to how the diseased neurons sense and use oxygen. The researchers describe the technique and their findings in a research paper published online by the journal JNeurosci. "Even a small portion of the brain can have dozens of different cell types," said Andreas Pfenning, an assistant professor in CMU's Computational Biology Department. "Each of these cell types has different roles in the behavior of an animal and also in disease." Separating cells of a certain type from their neighbors is thus a critical first step for researchers who want to study them. In this case, the research team focused on parvalbumin-expressing (PV+) neurons, which have been implicated in Parkinson's disease by the lab of Aryn Gittis, associate professor of biological sciences. Mice with Parkinson's symptoms regain motor control and their ability to run around when these cells are stimulated. Lab mice have been bred with PV+ cells that contain a protein called Cre that activates a fluorescent green protein. That fluorescence makes it possible for cell-sorting machines to isolate the cells from others in a mixture. But cell-sorting machines are extremely expensive, so Lawler developed a cheaper method, called Cre-Specific Nuclear Anchored Independent Labeling, or cSNAIL. The new technique uses a virus commonly employed by researchers to deliver DNA to brain cells. When the virus enters PV+ cells, Cre causes the tag to fluoresce. In the case of cSNAIL, researchers use antibodies to detect the tag and pull the PV+ nuclei away from other cells. "The technique turned out to be really specific, really efficient," Lawler said, noting that it can be adapted to other mouse models that use the Cre protein. In a subsequent analysis of the PV+ neurons, the researchers found that those from sick mice produced more RNA involved in the expression of genes that sense or use oxygen. Further study also showed that the DNA in the nucleus unwound in ways indicating that the oxygen-sensing genes were more active. "Oxygen-sensing pathways have been implicated in other, earlier aspects of Parkinson's disease, but not previously in PV+ cells," Lawler said. These pathways are involved in both protecting and killing cells during neurodegeneration. Pfenning noted that datasets from this study are part of a larger effort to build machine learning models that will help researchers interpret disease mechanisms by looking at how particular DNA sequences respond to different conditions across types of cells. "We're learning how to talk to cells, to speak their language," Lawler said. The National Institutes of Health and the National Science Foundation supported this research.

World's Fastest Open-Source Intrusion Detection Arrives

Intrusion-detection systems are the invisible intelligence agencies in computer networks. They scan every packet of data passed through the network, looking for signs of any one of the tens of thousands of cyberattack styles they recognize. As internet speeds increase, data volumes grow. To keep up, intrusion-detection systems have morphed into giant racks and stacks of servers, driving up energy costs for organizations that rely on them. That's all about to change. Researchers in Carnegie Mellon University's CyLab have developed the fastest-ever open-source intrusion-detection system — one that achieves speeds of 100 gigabits per second using a single server. "What was previously possible with 100-700 processor cores and a whole rack of machines, we can now do with five processor cores in a single server," said CyLab's Justine Sherry, an assistant professor in School of Computer Science's Computer Science Department. The researchers presented their work at the recent USENIX Symposium on Operating Systems Design and Implementation. Key to the researchers' success is using a field-programmable gate array (FPGA), an integrated circuit that users can program with customized code. The researchers programmed the FPGA specifically to detect intrusion, employing algorithms that are significantly faster than previous ones and that could not run on traditional processors. Sherry said that the FPGA processes an average of 95% of data packets on its own when it's placed in a network. The other 5% continue to central processing units when the FPGA becomes overwhelmed, hence the system's five processor cores. "The FPGA does most of the work, but some of it still goes to the processors," Sherry said. The new system produces enormous energy savings. To do the same work as the FPGA, a traditional system comprising hundreds of processing cores would use 38 times more power. "It's like your electricity bill used to be $100, and now it's $3," said Sherry. "We created one pizza box-sized machine to do the work of a whole room of servers." The researchers' open-source code can be downloaded on GitHub.

Zhu, Kaufman Receive Chairs Honoring Longtime HCII Faculty

Haiyi Zhu and Geoff Kaufman, both faculty members in the Human-Computer Interaction Institute (HCII), are the inaugural recipients of junior professorships funded by a gift from HCII Professor Jason Hong and his wife, Shelley Zhang. The professorships are named in honor of two prominent, longtime HCII faculty members: Daniel Siewiorek and Robert Kraut. Zhu is the Daniel P. Siewiorek Assistant Professor of Human-Computer Interaction and Kaufman is the Robert E. Kraut Assistant Professor of Human-Computer Interaction. They are among 13 newly appointed career development chair recipients across the university who will be honored during a virtual celebration at 5 p.m. on Wednesday, Nov. 11. Both also will be virtually feted during a pre-reception at 4 p.m. that will also feature Hong and HCII Director Jodi Forlizzi. Zhu is a social computing researcher who joined the HCII last year. Her broad interests are in the design and social impact of AI technologies in online and offline communities. She is particularly interested in developing AI systems that respect and balance community stakeholders' values. For instance, she has studied how algorithmic tools Wikipedia uses to judge the quality of edits and take corrective action can better align with the values of the Wikipedia community. She earned her undergraduate degree in computer science at Tsinghua University and her master's degree and Ph.D. at the HCII. She served as an assistant professor at the University of Minnesota, Twin Cities, before joining the HCII faculty.

Fragkiadaki Wins Air Force Young Investigator Award

Katerina Fragkiadaki, an assistant professor in the Machine Learning Department, is one of 36 scientists and engineers nationwide — and one of just two from Carnegie Mellon University — to receive funding this year through the Air Force's Young Investigator Research Program (YIP). Giulia Fanti, an assistant professor in the Electrical and Computer Engineering Department who has a courtesy appointment in the Computer Science Department, also won a YIP grant. The YIP aims to foster creative basic research in science and engineering, enhance early career development of outstanding young investigators, and increase opportunities for young investigators to recognize the Air Force mission and related challenges in science and engineering. Recipients receive a three-year grant totaling $450,000. The award will support Fragkiadaki's efforts to develop multimodal perception systems that use lifelong learning to improve their performance with little human guidance. This continual-learning framework would acquire common-sense knowledge about the world through agents embodied in simulated and real-world scenes. The agents would sense, move in and interact with their environments. They also would interact with human analysts who narrate and explain events in natural language. Fragkiadaki earned her bachelor's degree at the National Technical University of Athens, and her Ph.D. at the University of Pennsylvania. Before joining CMU, she was a post-doctoral researcher at the University of California, Berkeley, and at Google Research. Earlier this year, Fragkiadaki received the National Science Foundation's prestigious Faculty Early Career Development (CAREER) Award.

CMU Helps Provide Free WiFi to Area Students

Partnership With Pitt, Community Groups Creates Unique Nonprofit Network

Carnegie Mellon University has joined with Meta Mesh Wireless Communities, the University of Pittsburgh and the Keystone Initiative for Network-Based Research (KINBER) in an unprecedented partnership to provide free, high-speed internet access primarily to school-age children as the COVID-19 pandemic forces the integration of remote instruction. The new program, called Every1online, will provide WiFi to eligible households during a 12-month pilot program and has begun enrolling households in the Coraopolis, Homewood and New Kensington areas. The goal is to expand the program to additional communities in Western Pennsylvania after the pilot period. "To live and work in today's digital society, it is essential to have access to the internet," said Adam Longwill, executive director of Meta Mesh, a nonprofit organization that will construct and operate the WiFi network. "Unfortunately, many people in our own city of Pittsburgh cannot afford the internet services they need. Every1online is one way to begin to see internet access as a human right." Every1online is the first nonprofit wireless internet model in the country. Collaborators from the CMU School of Computer Science and the Simon Initiative are providing critical start-up funding for the project and serving as liaisons with key stakeholders and community groups. They will provide ongoing project design, implementation and fundraising support. "This kind of interorganizational collaboration, where large institutions and community organizations come together to solve complex, pressing challenges together, is essential for developing meaningful, impactful solutions," said Maggie Hannan, associate director of the Simon Initiative. "We are so thrilled to be a part of this collaborative and to have this chance to make real change in our city." Every1online will provide high-speed broadband internet that is optimal for simultaneous video conferences. While the pilot program seeks to prioritize households with students and those in need, any household within the service area can sign up to receive in-home equipment and connect to the network. Signups will continue through Jan. 31, 2021. Visit Meta Mesh's project page for more details. "So many of our students live in poverty, or other circumstances that prevent them from having access to high-speed internet, which is what they need to learn remotely," said Kris Hupp, director of technology and instructional innovation for the Cornell School District in Coraopolis. "This collaboration is really making the statement that these kids deserve equitable internet access." Ashley Williams Patton, director of the Computer Science Pathways program in SCS, emphasized that the goal isn't just to provide a Band-Aid to help students during the pandemic but to also address ongoing issues relating to internet access. "COVID-19 did not create the digital divide, but it is highlighting existing inequities in the education system," she said. "We're attempting to create a solution that doesn't go away — that sustainably alleviates a systemic problem." Pitt's Cathedral of Learning will serve as a super node/hub for the network. In addition to funding, Pitt is providing its extensive community engagement efforts and will support stakeholder and community involvement in additional neighborhoods following the pilot. KINBER, Pennsylvania's statewide research, education, and community network organization, will provide a gateway to the internet through its PennREN Fiber. The New Kensington-Arnold and Cornell school districts, as well as the Homewood Children's Village, are collaborators. Every1oneline is made possible with support from the Hopper Dean Foundation, the R.K. Mellon Foundation and the Burrell-New Kensington Rotary Club.

Waibel Team's AI Outperforms Humans in Speech Recognition

Alex Waibel, who holds faculty appointments in both the Language Technologies Institute and at the Karlsruhe Institute of Technology (KIT) in Germany, reports that his German lab has developed a computer system that for the first time outperforms people in recognizing conversational speech. It's difficult even for people to accurately transcribe conversations, Waibel said. "When people talk to each other, there are stops, stutters, hesitations such as 'er' or 'hmmm,' laughs, and coughs," he said. "Often, words are pronounced unclearly." With humans capable of no better than a 5% error rate, the feat has remained a major challenge for artificial intelligence. Now, KIT scientists and the staff of a KIT startup company called KITES have developed a system that boasts a 5% error rate with just a one-second delay, or latency. The measurements are based on an internationally recognized benchmark for the task established by the Defense Advanced Research Projects Agency. Waibel said fast, high accuracy speech recognition will enable better voice-based interactions with machines.

Beyond Data

CREATE Lab Launches Project That Helps Renters Avoid Evictions

Data is at the heart of EarthTime, a visualization tool developed by the Robotics Institute's CREATE Lab, and Pittsburgh housing data has been a particular focus for project scientist Anne Wright for the past two years. That focus shifted, however, with the onset of the coronavirus pandemic this past spring. Now, Wright and her colleagues scrape the local court websites every day, gathering information about eviction filings and hearings. As part of an initiative called Eviction Rapid Response, they work with a variety of local organizations and volunteers to act on the data by alerting tenants about their rights and linking them to programs that can help them stay in their homes. "People are being thrown into the deep end of a pool they didn't know existed," Wright said, noting that many tenants facing financial hard times during the COVID-19 pandemic haven't previously been through the eviction process. Tenants often don't know what they can do to halt or delay the loss of their dwelling. Illah Nourbakhsh, K&L Gates Professor of Ethics and Computational Technologies and director of the CREATE Lab, said the lab has always worked closely with community groups. But the role of its researchers necessarily expanded in response to the COVID-19 pandemic. "When COVID-19 hit, we focused on the urgent needs of the community," he said. That meant making the lab's unique resources and capabilities — such as data-gathering and logistics — available to local groups to fill gaps in services. "This is very different from operating in an ivory tower," he noted. "I think what we're doing is unique." The EarthTime housing project has addressed questions such as racial equity in mortgage approvals, effects of gentrification and how families are displaced across school attendance areas. Some of this data is readily available from public sources, such as the federal Housing Mortgage Disclosure Act, which the researchers then map out. Eviction data, Wright said, has been harder to come by. Initially, the team depended on a previous study by a Philadelphia researcher for information on evictions through 2018. Gathering up-to-date information became possible only this spring — during Allegheny County's eviction moratorium — thanks to a program Wright developed with help from the CREATE Lab's Randy Sargent and Amy Gottsegen. "I could see the evictions in real-time, but I couldn't do much about it," she recalled, which led her to launch Eviction Rapid Response. With funding from the Heinz Endowments, the CREATE Lab works with such organizations as the Allegheny County CARES Rent Relief Program run by Action Housing, the RentHelpPGH resource navigation project and the Allegheny County Homelessness Assistance Program. "The first thing we do is make sure they know they're being evicted," Wright said. "Then we let them know that, okay, so the landlord taped something to your door — that doesn't mean you've been evicted. It's just the first step in a long process." The project team mails flyers to renters to let them know about their rights and the resources available to them, and they collaborate with the public aid lawyer community to provide legal help. They also coordinate with a network of volunteers to encourage renters to go to their eviction hearings — otherwise, they'll be out on the street — and to distribute educational materials at court offices when eviction hearings are on the docket. About 14,000 evictions occur each year in Allegheny County. Despite the financial stress of the pandemic, the number of eviction filings are only around 3,400 as of late October, thanks to county and state moratoriums (that have since expired). A Centers for Disease Control and Prevention eviction protection program now provides some relief through the end of the year. Nevertheless, the pandemic has raised awareness and concern about evictions, Wright said. "People now understand that evicting people risks killing them," she said. "That's always been true, but now it's top of mind."